C++ API runtime type selection

Hello, this is just a question for understanding :-)

I have a c++ mex function that takes in a real scalar. In the function code I have the assignment

double x=inputs[0][0];

This does the right conversion even when the MATLAB input type is not double, e.g. single or uint8.

As I understand it, the way this works is that input[0] is a TypedArray<T>. Depending on T, the assignment double x=input[0][0]; figures out at compile time how the bits in the input are converted to the bits of a double. And the internal c++ code of the MATLAB c++ api would need to look like

"If the MATLAB input is of type uint8, create a TypedArray<uint8_t>.

if the MATLAB input type is single, create TypedArray<float>.

etc."

But templates are a compile time construct, so how is that even possible?

[James Tursa]

Please post a minimum working example of code that runs and then we can comment on how it works or doesn't work.

[Klaus Ebenhauer]

I'm sorry for answering so late. Here is a minimum working example: The file "mySquare.cpp" contains

#include "mex.hpp"
#include "mexAdapter.hpp"

using matlab::mex::ArgumentList;
using namespace matlab::data;

class MexFunction : public matlab::mex::Function {
public:
	void operator()(ArgumentList outputs, ArgumentList inputs) {
		double x = inputs[0][0];
		ArrayFactory factory;
		TypedArray<double> t = factory.createScalar(x * x);
		outputs[0] = t;
	}
};

which I compiled in Matlab with

mex mySquare.cpp

(with MinGW 6.3.0 on Win10 64bit), creating "mySquare.mexw64". Then, both

mySquare(3)

and e.g.

mySquare(uint8(3))

return the correct result double(9).